Plasma processing apparatus, control method and program
The plasma processing apparatus controls the EEDF by measuring and adjusting parameters in real-time using a learning model, optimizing plasma processing efficiency and stability without requiring additional experimental validation.
Patent Information
- Application Number
- JP2023548413
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2021-09-17
- Filing Date
- 2022-09-05
- Publication Date
- 2024-11-06
- Estimated Expiration
- 2042-09-05
AI Technical Summary
Existing plasma processing technologies struggle to accurately control the electron energy distribution function (EEDF) within a processing chamber, which varies based on process recipes and chamber conditions, hindering optimal plasma treatment.
A plasma processing apparatus equipped with a control device that measures the EEDF during processing and adjusts parameters to align it with a target EEDF using a learning model that updates in real-time, employing Bayes' theorem to optimize the control law.
This approach allows for precise control of the EEDF, enabling more efficient and stable plasma processing by continuously learning and adapting to changes, reducing the need for experimental adjustments and minimizing resource waste.
Smart Images

Figure 0007578381000007 
Figure 0007578381000008 
Figure 0007578381000009
Abstract
Description
[Technical field]
[0001] The present disclosure relates to a plasma processing apparatus, a control method, and a program. [Background technology]
[0002] For example, in a plasma processing apparatus, it is known that the electron energy distribution function (EEDF) of the plasma in the chamber varies depending on the process recipe, the chamber condition, etc. When optimizing the plasma processing, it is desirable to make the electron energy distribution function of the plasma in the chamber, for example, closer to an ideal shape.
[0003] Information processing devices that input start state data of a processed object and end state data of a target processed object to determine predetermined processing conditions for a process performed on the processed object have been known for some time. It has been disclosed that the conventional information processing devices perform machine learning to improve the efficiency of searching for an optimal solution. In addition, plasma density is disclosed as an example of data related to a semiconductor manufacturing process (see, for example, Patent Document 1). [Prior art documents] [Patent documents]
[0004] [Patent Document 1] International Publication No. 2019 / 155928 Summary of the Invention [Problem to be solved by the invention]
[0005] The present disclosure provides a technique for controlling the electron energy distribution function of plasma in a processing vessel to approach a target electron energy distribution function by changing parameters related to plasma processing. [Means for solving the problem]
[0006] One aspect of the present disclosure is a plasma processing apparatus having a control device that converts a gas supplied into a processing vessel into plasma and performs plasma processing on a workpiece, the control device having a measurement unit that measures an electron energy distribution function of the plasma in the processing vessel, and a parameter modification unit that modifies parameters related to the plasma processing so that the electron energy distribution function measured by the measurement unit during plasma processing approaches a target electron energy distribution function. Effect of the Invention
[0007] According to the present disclosure, it is possible to provide a technique for controlling the electron energy distribution function of plasma in a processing chamber to approach a target electron energy distribution function by changing parameters related to plasma processing. [Brief description of the drawings]
[0008] [Figure 1] 1 shows an example of a cross-sectional view of a plasma processing apparatus according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram illustrating a hardware configuration of an example of a computer. [Diagram 3] FIG. 2 is a diagram illustrating an example of a functional configuration of a control device according to the present embodiment. [Figure 4] 4 is a flowchart showing an example of a processing procedure of the plasma processing apparatus according to the present embodiment. [Diagram 5] FIG. 2 is an explanatory diagram of an example of an ideal electron energy distribution function of plasma. [Figure 6A] FIG. 4 is an explanatory diagram of an example of an evaluation function. [Figure 6B] FIG. 4 is an explanatory diagram of an example of an evaluation function. [Figure 6C] FIG. 4 is an explanatory diagram of an example of an evaluation function. [Figure 6D] FIG. 4 is an explanatory diagram of an example of an evaluation function. [Figure 7] FIG. 1 is a diagram showing an example of a probability distribution (prior distribution) of the values of each coefficient in a learning model. [Figure 8]1A and 1B are diagrams illustrating an example of a process for measuring an electron energy distribution function of plasma in a processing vessel during plasma processing; [Figure 9] FIG. 11 is an explanatory diagram of an example of a process parameter changing process; [Figure 10A] 11A and 11B are diagrams illustrating an example of the results of evaluation of an electron energy distribution function of plasma during plasma processing, which changes with changes in process parameters. [Figure 10B] 11A and 11B are diagrams illustrating an example of the results of evaluation of an electron energy distribution function of plasma during plasma processing, which changes with changes in process parameters. [Figure 11] 11A and 11B are diagrams illustrating an example of the results of evaluation of an electron energy distribution function of plasma during plasma processing, which changes with changes in process parameters. [Figure 12] FIG. 4 is an explanatory diagram of an example of updating a control law. [Figure 13] FIG. 11 is an explanatory diagram of an example of updating the probability distribution (prior distribution) of the values of each coefficient in a learning model that defines a control law. [Figure 14A] FIG. 11 is an explanatory diagram of an example of updating the probability distribution (prior distribution) of the values of each coefficient in a learning model that defines a control law. [Figure 14B] FIG. 11 is an explanatory diagram of an example of updating the probability distribution (prior distribution) of the values of each coefficient in a learning model that defines a control law. [Figure 14C] FIG. 11 is an explanatory diagram of an example of updating the probability distribution (prior distribution) of the values of each coefficient in a learning model that defines a control law. [Figure 15A] 11 is an explanatory diagram of an example of a process of repeating control of an electron energy distribution function of plasma by repeating learning of a control law. FIG. [Figure 15B] 11 is an explanatory diagram of an example of a process of repeating control of an electron energy distribution function of plasma by repeating learning of a control law. FIG. [Figure 15C] 11 is an explanatory diagram of an example of a process of repeating control of an electron energy distribution function of plasma by repeating learning of a control law. FIG. [Figure 15D] 11 is an explanatory diagram of an example of a process of repeating control of an electron energy distribution function of plasma by repeating learning of a control law. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Note that in this specification and the drawings, illustrations and descriptions of parts that are not necessary for the description of the present embodiment will be omitted as appropriate.
[0010] [Plasma processing equipment] 1 shows an example of a cross-sectional view of a plasma processing apparatus according to an embodiment of the present invention. The plasma processing apparatus 100 has a processing vessel (chamber) 1 that accommodates a wafer W. The plasma processing apparatus 100 performs plasma processing on the wafer W by using surface wave plasma formed on the inner wall surface of the ceiling of the processing vessel 1 by microwaves. The plasma processing is a film formation process, an etching process, an ashing process, or the like.
[0011] The plasma processing apparatus 100 includes a processing vessel 1, a microwave plasma source 2, and a control apparatus 3. The processing vessel 1 is an airtight, generally cylindrical vessel made of a metal material such as aluminum or stainless steel, and is grounded.
[0012] The processing vessel 1 has a main body 10 and forms a plasma processing space therein. The main body 10 is a disk-shaped top plate that constitutes the ceiling of the processing vessel 1. A support ring 129 is provided on the contact surface between the processing vessel 1 and the main body 10. The inside of the processing vessel 1 is hermetically sealed. The main body 10 is made of a metal material such as aluminum or stainless steel.
[0013] The microwave plasma source 2 has a microwave output unit 30, a microwave transmission unit 40, and a microwave radiation mechanism 50. The microwave output unit 30 outputs microwaves by distributing them to multiple paths. The microwaves are introduced into the processing vessel 1 through the microwave transmission unit 40 and the microwave radiation mechanism 50. The gas supplied into the processing vessel 1 is excited by the electric field of the introduced microwaves, and as a result, a surface wave plasma is formed.
[0014] A mounting table 11 on which a wafer W is placed is provided within the processing vessel 1. The mounting table 11 is supported by a cylindrical support member 12 that is erected via an insulating member 12a at the center of the bottom of the processing vessel 1. Examples of materials that may be used to form the mounting table 11 and the support member 12 include metals such as aluminum whose surfaces have been anodized (anodized), and insulating members (ceramics, etc.) having high-frequency electrodes therein. The mounting table 11 may be provided with an electrostatic chuck for electrostatically adsorbing the wafer W, a temperature control mechanism, a gas flow path for supplying a gas for heat transfer to the rear surface of the wafer W, and the like.
[0015] A high frequency bias power supply 14 is connected to the mounting table 11 via a matching box 13. When high frequency power is supplied from the high frequency bias power supply 14 to the mounting table 11, ions in the plasma are attracted to the wafer W. Note that the high frequency bias power supply 14 may not be provided depending on the characteristics of the plasma processing.
[0016] An exhaust port is provided at the bottom of the processing vessel 1. An exhaust pipe 15 is connected to the exhaust port. The exhaust pipe 15 is connected to an exhaust device 152 via a pressure control valve 151. The exhaust pipe 15, the pressure control valve 151, and the exhaust device 152 form a gas exhaust section 16. The exhaust device 152 has a vacuum pump such as a turbo molecular pump.
[0017] When the gas exhaust unit 16 is operated, the inside of the processing vessel 1 is exhausted and the pressure is quickly reduced to a predetermined vacuum level. A pressure gauge 160 is also installed in the processing vessel 1. The pressure gauge 160 measures the pressure value inside the processing vessel 1. The measurement result is received by the control device 3. The pressure control valve 151 controls its opening based on the measured pressure value.
[0018] The side wall of the processing vessel 1 is provided with a loading / unloading port 17 for loading / unloading the wafer W, and a gate valve 18 for opening / closing the loading / unloading port 17. The microwave transmission unit 40 transmits the microwaves output from the microwave output unit 30. The central microwave introduction unit 43b in the microwave transmission unit 40 is disposed at the center of the main body unit 10. The six peripheral microwave introduction units 43a are disposed at equal intervals in the circumferential direction around the main body unit 10. The central microwave introduction unit 43b and the six peripheral microwave introduction units 43a have a function of introducing the microwaves output from the amplifier units 42 shown in FIG. 1 corresponding to each of them to the microwave radiation mechanism 50 and a function of matching impedance. Hereinafter, the peripheral microwave introduction unit 43a and the central microwave introduction unit 43b are collectively referred to as the microwave introduction unit 43.
[0019] The six dielectric layers 123 are disposed below the six peripheral microwave introduction parts 43a and inside the main body part 10. Moreover, one dielectric layer 133 is disposed below the central microwave introduction part 43b and inside the main body part 10. The number of peripheral microwave introduction parts 43a and dielectric layers 123 is not limited to six, but may be two or more. However, the number of peripheral microwave introduction parts 43a and dielectric layers 123 is preferably three or more. The number of peripheral microwave introduction parts 43a and dielectric layers 123 may be, for example, three to six.
[0020] The microwave radiation mechanism 50 in Fig. 1 has a dielectric top plate 121, a dielectric top plate 131, a slot 122, a slot 132, a dielectric layer 123, and a dielectric layer 133. The dielectric top plates 121 and 131 are formed from a disk-shaped dielectric material that transmits microwaves, and are disposed on the upper surface of the main body 10. The dielectric top plates 121 and 131 are formed from a material having a relative dielectric constant greater than that of a vacuum, such as quartz, ceramics such as alumina (Al2O3), fluorine-based resins such as polytetrafluoroethylene, and polyimide-based resins. This makes it possible to make the wavelength of the microwaves that transmit through the dielectric top plates 121 and 131 shorter than the wavelength of the microwaves that propagate in a vacuum, thereby making it possible to make the antenna including the slots 122 and 132 smaller.
[0021] Under the dielectric top plates 121 and 131, dielectric layers 123 and 133 are in contact with the rear surface of the opening of the main body 10 via slots 122 and 132 formed in the main body 10. The dielectric layers 123 and 133 are made of, for example, quartz, ceramics such as alumina (Al2O3), fluorine-based resin such as polytetrafluoroethylene, or polyimide-based resin. The dielectric layers 123 and 133 are recessed from the ceiling surface by the thickness of the opening formed in the main body 10, and function as a dielectric window that supplies microwaves to the plasma generation space U.
[0022] The peripheral microwave introduction part 43a and the central microwave introduction part 43b are arranged coaxially with a cylindrical outer conductor 52 and a rod-shaped inner conductor 53 provided at the center of the cylindrical outer conductor 52. Microwave power is supplied between the outer conductor 52 and the inner conductor 53, forming a microwave transmission path 44 through which microwaves propagate toward the microwave radiation mechanism 50.
[0023] The peripheral microwave introduction part 43a and the central microwave introduction part 43b are provided with a slug 54 and an impedance adjustment member 140 located at the tip of the slug 54. By moving the slug 54, the impedance of the load (plasma) in the processing vessel 1 can be matched to the characteristic impedance of the microwave power source in the microwave output part 30. The impedance adjustment member 140 is made of a dielectric material and adjusts the impedance of the microwave transmission line 44 by its relative dielectric constant.
[0024] The main body 10 is provided with a gas inlet 21 having a shower structure. Gas supplied from a gas supply source 22 passes through the gas inlet 21 from a gas diffusion chamber 62 via a gas supply pipe 111, and is supplied in a shower-like manner into the processing vessel 1. The gas inlet 21 is, for example, an example of a gas shower head that supplies gas from a plurality of gas supply holes 60 formed in the ceiling wall of the processing vessel 1. Examples of the gas include a gas for plasma generation such as Ar gas, a gas to be decomposed with high energy such as O2 gas or N2 gas, and a processing gas such as silane gas.
[0025] An opening 1b is formed in the sidewall of the processing vessel 1 in the circumferential direction, and a plasma probe device 70 is attached to the opening 1b. One or more plasma probe devices 70 may be attached to the processing vessel 1. The plasma probe device 70 senses the plasma generated in the plasma generation space U.
[0026] The plasma probe device 70 is connected to a monitor device 80. The monitor device 80 has a signal generator and outputs a signal of a predetermined frequency transmitted by the signal generator. The output signal of the predetermined frequency is transmitted to the plasma probe device 70 by a coaxial cable 81, and is transmitted to the plasma from the antenna unit 71 at the tip of the plasma probe device 70.
[0027] The plasma probe device 70 detects a current value from the plasma side in response to a signal transmitted to the plasma side, and sends it to the monitor device 80. The detected current value is transmitted from the monitor device 80 to the control device 3. The control device 3 measures the electron energy distribution function of the plasma in the processing vessel 1, for example, by FFT (frequency) analysis. In this way, the control device 3 measures the electron energy distribution function of the plasma in the processing vessel 1 from the frequency characteristics of the plasma generation space U obtained by electrical probe measurement. Details of the plasma probe device 70 and the monitor device 80 are described in, for example, JP 2019-46787 A.
[0028] Each part of the plasma processing apparatus 100 is controlled by a control device 3. The control device 3 controls each part of the plasma processing apparatus 100 based on a process sequence and a process recipe of the plasma processing apparatus 100. The control device 3 is also capable of inputting data and displaying results when performing a predetermined control according to the process sequence and the process recipe. The control device 3 may be built into the plasma processing apparatus 100, or may be connected to the plasma processing apparatus 100 via a communication path.
[0029] The communication path may be a wired communication system or a wireless communication system, and may be a signal path for exchanging various signals inside and outside the computer. The communication path may utilize a network such as a local area network (LAN).
[0030] When performing plasma processing in the plasma processing apparatus 100, the wafer W is carried into the processing vessel 1 through the open gate valve 18 and the loading / unloading port 17 while being held on the transfer arm. The gate valve 18 is closed after the wafer W is carried in. When the wafer W is carried to above the mounting table 11, it is transferred from the transfer arm to the pusher pin, and the pusher pin descends to place the wafer W on the mounting table 11. The pressure inside the processing vessel 1 is maintained at a predetermined vacuum level by the gas exhaust unit 16. Gas is introduced into the processing vessel 1 in a shower-like manner from the gas introduction unit 21. Microwaves radiated from the microwave radiation mechanism 50 via the peripheral microwave introduction unit 43a and the central microwave introduction unit 43b propagate along the inner surface of the ceiling wall. The gas is excited by the electric field of the microwaves propagating as surface waves. The wafer W is subjected to plasma processing by the surface wave plasma generated in the plasma generation space U below the ceiling wall on the processing vessel 1 side.
[0031] The control device 3 is realized by, for example, the computer 500 shown in Fig. 2. The control device 3 reads a program recorded in a storage device, and performs plasma processing by transmitting and receiving various signals to and from each part constituting the plasma processing device 100 in accordance with the program.
[0032] The control device 3 is realized by, for example, a computer having a hardware configuration as shown in Fig. 2. Fig. 2 is a diagram showing an example of the hardware configuration of a computer.
[0033] 2 includes an input device 501, an output device 502, an external I / F (interface) 503, a RAM (Random Access Memory) 504, a ROM (Read Only Memory) 505, a CPU (Central Processing Unit) 506, a communication I / F 507, and a HDD (Hard Disk Drive) 508, all of which are connected to each other via a bus B. The input device 501 and the output device 502 may be connected and used when necessary.
[0034] The input device 501 is a keyboard, mouse, touch panel, etc., and is used by an operator to input various operation signals. The output device 502 is a display, etc., and displays the results of processing by the computer 500. The communication I / F 507 is an interface that connects the computer 500 to a network, etc. The HDD 508 is an example of a non-volatile storage device that stores programs and data.
[0035] The external I / F 503 is an interface with an external device. The computer 500 can read and / or write data from and to a recording medium 503a such as a Secure Digital (SD) memory card via the external I / F 503. The ROM 505 is an example of a non-volatile semiconductor memory (storage device) in which programs and data are stored. The RAM 504 is an example of a volatile semiconductor memory (storage device) that temporarily holds programs and data.
[0036] The CPU 506 is a calculation device that reads out programs and data from storage devices such as the ROM 505 and the HDD 508 onto the RAM 504 and executes processes, thereby realizing overall control and functions of the computer 500 .
[0037] The control device 3 shown in FIG. 1 can realize various functions shown in FIG. 3 by the computer 500 having the hardware configuration shown in FIG. 2 executing processes according to a program.
[0038] Fig. 3 is a diagram showing an example of the functional configuration of the control device according to the present embodiment. The control device 3 shown in Fig. 3 includes a target setting reception unit 200, an evaluation function setting reception unit 202, a control law setting reception unit 204, a plasma processing control unit 206, an electron energy distribution function measurement unit 208, a parameter change unit 210, a control law update unit 212, a target storage unit 220, an evaluation function storage unit 222, a control law storage unit 224, and a process recipe storage unit 226.
[0039] The target setting receiving unit 200 receives the setting of a target plasma electron energy distribution function from an operator, etc. The target plasma electron energy distribution function is an electron energy distribution function of plasma that is considered ideal in a plasma processing of interest.
[0040] The ideal electron energy distribution function of the plasma can be determined by experiments, calculations, literature surveys, etc. For example, the ideal electron energy distribution function of the plasma can be determined based on the purpose of the plasma processing and the chemical reactions in the plasma, etc. Also, the ideal electron energy distribution of the plasma may be determined from the correlation between the measurement results of the electron energy distribution function of the plasma and the evaluation results of the process such as film quality.
[0041] In addition to the setting of an ideal electron energy distribution function of plasma corresponding to the plasma processing of interest, the target setting receiving unit 200 may also receive a setting of an ideal electron energy distribution function of plasma corresponding to a location in the processing vessel 1. Furthermore, the target setting receiving unit 200 may also receive a setting of an ideal electron energy distribution function of plasma that does not change with the passage of time in the plasma processing, or may receive a setting of an ideal electron energy distribution function of plasma that changes with the passage of time in the plasma processing.
[0042] For example, an ideal electron energy distribution function of a plasma can be defined by the following formula (1). In formula (1), x is an example of a factor that determines the shape of the electron energy distribution function. T eff is an example of a factor that determines the spread of the electron energy distribution function.
[0043]
number
[0044] In addition, the evaluation function setting receiving unit 202 receives, from an operator or the like, the setting of an evaluation function that quantifies the similarity or discrepancy between the electron energy distribution function of the plasma measured during plasma processing and the electron energy distribution function of the target plasma.
[0045] For example, an evaluation function that quantifies the difference between the electron energy distribution function of the plasma measured during plasma processing and the electron energy distribution function of the target plasma can be defined by the following formula (2). In formula (2), exp represents the measurement result, and ideal represents the ideal form.
[0046]
number
[0047] For example, formula (2) can be expressed as x and T eff This is an example of an evaluation function that quantifies the deviation between the electron energy distribution function of the plasma measured during plasma processing and the electron energy distribution function of the target plasma, using vector h defined by the difference between the measurement result of and the ideal form. For example, since it is considered that the quality of plasma differs significantly depending on whether there is an excess of high-energy electrons or an excess of low-energy electrons, it is considered that the evaluation function of formula (2) is more desirable than the evaluation function of the sum of squares of the residuals.
[0048] In addition, the control law setting reception unit 204 receives settings of an initial learning model that represents the correspondence between parameters related to the plasma processing and the results of the evaluation function from an operator, etc. For example, the control law setting reception unit 204 receives settings of M process parameters {θ1, θ2, ..., θ M The initial training model h is represented by vector h by i =Φ i (θ1, θ2, …, θ M ) settings are accepted.
[0049] The process parameters are an example of parameters related to plasma processing, such as the input power, the frequency of the input power, the duty ratio of the input power pulse, the gas mixture ratio, the pressure, the temperature, the cumulative number of processes, the most recent process recipe, the most recently measured electron energy distribution function, etc. The shape of the learning model and each coefficient in the learning model define the control law.
[0050] When there are two process parameters (M=2), the control law setting receiving unit 204 can receive the setting of a learning model such as the following formula (3), for example.
[0051]
number
[0052] The learning model for which the control law setting reception unit 204 accepts settings can be determined by experiments, calculations, literature research, and the like. The learning model for which the control law setting reception unit 204 accepts settings is an initial learning model, which is updated to learn and improve accuracy by repeating plasma processing as described later. The control law setting reception unit 204 also expresses the probability distribution (prior distribution) of the values of each coefficient in the learning model that defines the control law using a range estimate as described later. The range estimate is expressed, for example, by a normal distribution centered on a point estimate. The probability distribution of the values of each coefficient in the learning model that defines the control law can be expressed, for example, by drawing shading of probability p(a) in a space containing vector a as shown in equation (4).
[0053]
number
[0054] The plasma processing control unit 206 controls the plasma processing apparatus 100 so that plasma processing is performed according to the process parameters. The electron energy distribution function measurement unit 208 measures the electron energy distribution function of the plasma in the processing vessel 1 during plasma processing. The electron energy distribution function measurement unit 208 can measure the electron energy distribution function of the plasma by an existing insulated probe method that is applicable to plasma processes.
[0055] For example, the electron energy distribution function measuring unit 208 can derive the electron energy distribution function of the plasma in the processing vessel 1 from the frequency characteristics of the plasma generating space U obtained by electrical probe measurement.
[0056] Furthermore, the parameter changing unit 210 evaluates the plasma electron energy distribution function, which is the measurement result of the electron energy distribution function measuring unit 208, in accordance with the evaluation function set in the evaluation function setting receiving unit 202. For example, the parameter changing unit 210 quantifies the deviation between the plasma electron energy distribution function, which is the measurement result, and the target plasma electron energy distribution function set in the evaluation function setting receiving unit 202.
[0057] The parameter change unit 210 changes the process parameters according to a learning model that represents the correspondence between the process parameters and the evaluation function results, so that the plasma electron energy distribution function, which is the measurement result, approaches the target plasma electron energy distribution function. The process parameters may include parameters that can be changed (controlled) and parameters that cannot be changed. The input power, the frequency of the input power, the duty ratio of the input power pulse, the gas mixture ratio, the pressure, and the temperature are examples of parameters that can be changed. The cumulative number of processes, the most recent process recipe, and the most recently measured electron energy distribution function are examples of parameters that cannot be changed.
[0058] The control law update unit 212 updates the learning model based on the evaluation result of the electron energy distribution function of the plasma during plasma processing with the process parameters changed by the parameter change unit 210. The control law update unit 212 updates each coefficient in the learning model using the response of the electron energy distribution function of the plasma due to the change in the process parameters as new information source.
[0059] If the result of the evaluation of the electron energy distribution function of the plasma during plasma processing using the process parameters changed by the parameter change unit 210 is not as expected, the control law update unit 212 determines that the accuracy of the learning model is poor (the control law is incorrect). Furthermore, if the result of the evaluation of the electron energy distribution function of the plasma during plasma processing using the process parameters changed by the parameter change unit 210 is as expected, the control law update unit 212 determines that the accuracy of the learning model is good (the control law is correct). If the control law update unit 212 determines that the accuracy of the learning model is poor, it updates each coefficient in the learning model using the response of the electron energy distribution function of the plasma due to the change in the process parameters as a new information source. Furthermore, the control law update unit 212 updates the probability distribution of the value of each coefficient in the learning model that defines the control law.
[0060] During plasma processing, the controller 3 repeatedly changes the process parameters by the parameter changing unit 210 and updates the learning model by the control law updating unit 212 so as to approach a target plasma electron energy distribution function. This improves the accuracy of the controller 3's control of the plasma electron energy distribution function, allowing the plasma electron energy distribution function to approach an ideal form.
[0061] The target memory unit 220 stores an electron energy distribution function of a target plasma whose setting is accepted by the target setting accepting unit 200. The evaluation function memory unit 222 stores an evaluation function whose setting is accepted by the evaluation function setting accepting unit 202. The control law memory unit 224 stores an initial learning model whose setting is accepted by the control law setting accepting unit 204, and a learning model updated by the control law update unit 212. The process recipe memory unit 226 stores a process sequence and a process recipe.
[0062] [process] Fig. 4 is a flowchart showing an example of a processing procedure of the plasma processing apparatus according to this embodiment. The flowchart in Fig. 4 shows a process of controlling the electron energy distribution function of plasma, which is executed by the control device 3, among the processes of the plasma processing apparatus 100.
[0063] In step S10, the target setting receiving unit 200 of the control device 3 receives a setting of a target plasma electron energy distribution function defined by, for example, the above-mentioned formula (1) from an operator, etc. The target setting receiving unit 200 may receive a setting of a target plasma electron energy distribution function as shown in FIG.
[0064] Fig. 5 is an explanatory diagram of an example of an ideal electron energy distribution function of plasma. The ideal electron energy distribution function of plasma in Fig. 5 is an example in which there are many electrons having energy capable of decomposing the material gas, and there are no electrons having energy lower than the energy capable of decomposing the material gas, and no electrons having energy higher than the energy capable of decomposing the material gas.
[0065] In step S12, the evaluation function setting receiving unit 202 receives the setting of an evaluation function defined, for example, as shown in Fig. 6A to Fig. 6D from an operator or the like. Fig. 6A to Fig. 6D are explanatory diagrams of an example of the evaluation function. The evaluation function in Fig. 6A to Fig. 6D quantifies the degree to which the electron energy distribution function of plasma measured during plasma processing (electron energy distribution function of the measurement result) is similar to or deviates from the target electron energy distribution function of plasma (electron energy distribution function of an ideal form).
[0066] FIG. 6A is an example of the electron energy distribution function of the measurement result and the ideal electron energy distribution function. FIG. 6B shows the difference between the electron energy distribution function of the measurement result of FIG. 6A and the ideal electron energy distribution function. FIG. 6C is an example of discretizing the difference of FIG. 6B. The fineness of discretization can be adjusted as appropriate, and does not necessarily have to be at equal intervals. FIG. 6D is an example of visually expressing the discretized difference vector h of FIG. 6C. Note that the discretized difference vector h may be made to express the part to be prioritized by weighting or the like. For example, in FIG. 6D, the part corresponding to electrons having high energy is multiplied by a constant in order to strictly evaluate electrons having high energy.
[0067] In step S14, the control law setting reception unit 204 receives settings of an initial learning model that represents the correspondence between the process parameters and the results of the evaluation function from an operator, etc. For example, the control law setting reception unit 204 receives settings of M process parameters {θ1, θ2, ..., θ M} to express the initial learning model h, for example, as shown in equation (5). i =Φ i (θ1, θ2, …, θ M ) settings are accepted.
[0068]
number
[0069] Furthermore, the control law setting receiving unit 204 expresses the probability distribution (prior distribution) of the values of each coefficient in the learning model that defines the control law by range estimates, for example, as shown in Fig. 7. Fig. 7 is a diagram showing an example of the probability distribution (prior distribution) of the values of each coefficient in the learning model. Fig. 7 shows an example of the prior distribution of coefficients a1 to a3 of the learning model shown in equation (3). If the number of coefficients is n, the probability distribution of the values of each coefficient in the learning model that defines the control law can be expressed by drawing shading of probability p(a) in an n-dimensional space in which vector a is located, for example, as shown in Fig. 7.
[0070] In step S16, the plasma processing control unit 206 controls the plasma processing apparatus 100 so that the plasma processing is performed according to the process parameters. The electron energy distribution function measurement unit 208 measures the electron energy distribution function of the plasma in the processing vessel 1 during the plasma processing at predetermined time intervals (e.g., every second).
[0071] Figure 8 is an explanatory diagram of an example of a process for measuring the electron energy distribution function of plasma in a processing vessel during plasma processing. The sheath rectification characteristic graph in Figure 8 is a graph of the relationship between the probe voltage and the probe current. The vertical axis represents the sum of the electron current and ion current flowing through the probe. By applying positive and negative voltages to the probe, the probe measures the electrons and ions in the plasma that are attracted to the probe as a current value.
[0072] The electron energy distribution function measurement unit 208 can derive the electron energy distribution function of the plasma in the processing vessel 1 from frequency characteristics obtained by FFT analysis of the probe current measured by the probe.
[0073] In step S18, when the measurement result of the electron energy distribution function of the plasma in the processing vessel 1 during plasma processing is acquired, the controller 3 performs the process of step S20. If the measurement result of the electron energy distribution function of the plasma in the processing vessel 1 during plasma processing is not acquired in step S18, the controller 3 skips the processes of steps S20 to S24 and performs the process of step S26.
[0074] In step S20, the parameter change unit 210 evaluates the electron energy distribution function of the plasma, which is the acquired measurement result, according to the evaluation function set in the evaluation function setting reception unit 202. The parameter change unit 210 obtains a process parameter θ' that becomes a vector h' of a minimum value, for example as shown in FIG. 9, according to a learning model that indicates the correspondence between the process parameter and the result of the evaluation function. FIG. 9 is an explanatory diagram of an example of a process for changing a process parameter. The parameter change unit 210 changes the process parameter from θ to θ'. The plasma processing control unit 206 controls the plasma processing apparatus 100 so that the plasma processing is performed according to the changed process parameter θ'.
[0075] For example, if the accuracy of the learning model is high, the evaluation result of the electron energy distribution function of the plasma during plasma processing with the process parameter θ' changed by the parameter changing unit 210 will be as expected, as shown in Figures 10A and 10B. Figures 10A and 10B are explanatory diagrams of an example of the evaluation result of the electron energy distribution function of the plasma during plasma processing that changes with the change of the process parameter.
[0076] On the other hand, if the accuracy of the learning model is low, the evaluation result of the electron energy distribution function of the plasma during plasma processing using the process parameter θ' changed by the parameter changing unit 210 will not be as expected, as shown in Fig. 11. Fig. 11 is an explanatory diagram of an example of the evaluation result of the electron energy distribution function of the plasma during plasma processing that changes due to the change of the process parameter. In Fig. 11, the evaluation result of the electron energy distribution function of the plasma during plasma processing using the changed process parameter θ' is not as expected, and the direction and norm of the vector h' are not as expected.
[0077] Therefore, the controller 3 updates the control law in steps S22 to S24 based on the result of evaluation of the electron energy distribution function of the plasma during plasma processing using the changed process parameter θ'.
[0078] In step S22, the control law update unit 212 determines whether or not the learning model needs to be updated based on the evaluation result of the electron energy distribution function of the plasma during the plasma processing using the process parameter θ' changed by the parameter change unit 210. For example, if the accuracy of the learning model is low, the control law update unit 212 determines that the learning model needs to be updated.
[0079] If it is determined that the learning model needs to be updated, the control law update unit 212 performs the process of step S24. If it is not determined that the learning model needs to be updated, the control law update unit 212 skips the process of step S24.
[0080] In step S24, the control law update unit 212 updates each coefficient in the learning model based on the evaluation result of the electron energy distribution function of the plasma during the plasma processing with the changed process parameter θ', for example, as shown in Fig. 12. Fig. 12 is an explanatory diagram of an example of updating the control law.
[0081] For example, even if the process parameter θ' is specified, there are multiple combinations of coefficients (vector a) in the learning model that can realize vector h' in Figure 12, as shown in Figure 12. For example, i =a i,1 +a i,2 When θ, (θ',h' i The combination that realizes )=(1,1) is a i,1 +a i,2 For all (a i,1 ,a i,2 ).
[0082] Therefore, in order to update to the most likely vector a, the control law update unit 212 updates the probability distribution (prior distribution) of the values of each coefficient in the learning model that defines the control law according to Bayes' theorem, as shown in the following equation (6).
[0083]
number
[0084] Figure 13 is an explanatory diagram of an example of updating the probability distribution (prior distribution) of the values of each coefficient in the learning model that defines the control law. Bayes' theorem is a method for finding the "probability P(A|B) of event A under the condition that a certain event B has occurred" using the "probability P(B|A) of event B under the condition that a certain event A has occurred."
[0085] In Figure 13, the probability distribution (prior distribution) of the values of each coefficient in the learning model, that is, the probability p(θ'∩h'|a) of the event θ'∩h' under the condition of event a, is used to calculate the probability distribution (posterior distribution) of the values of each coefficient in the learning model under the condition that the event θ'∩h' has occurred.
[0086] In this way, the control law update unit 212 redraws the probability distribution (prior distribution) of the value of each coefficient in the learning model that defines the control law into a posterior distribution based on the event θ'∩h'. For example, the most likely value may be selected from the posterior distribution for the value of each coefficient in the learning model that defines the control law.
[0087] According to the process of step S24, even if the evaluation result is not as expected as shown in Fig. 14B, the control law update unit 212 can update the probability distribution of the values of each coefficient in the learning model from the unexpect- ed event according to Bayes' theorem as shown in Fig. 14C. Figs. 14A to 14C are explanatory diagrams of an example of updating the probability distribution (prior distribution) of the values of each coefficient in the learning model that prescribes the control law.
[0088] The processing of steps S18 to S24 in Fig. 4 is repeated until the process ends. According to the processing of the flowchart shown in Fig. 4, the control device 3 can improve the accuracy of the learning model by repeating Fig. 15A to Fig. 15D. Fig. 15A to Fig. 15D are explanatory diagrams of an example of processing for repeating the learning of the control law and controlling the electron energy distribution function of the plasma.
[0089] FIG 15A shows the measurement of the electron energy distribution function of the plasma during plasma processing. FIG 15B shows the control of the plasma processing according to the modified process parameter θ'. FIG 15C shows the evaluation of the electron energy distribution function of the plasma during plasma processing with the modified process parameter θ'. FIG 15D shows the process of learning the values of each coefficient in the learning model by Bayes' theorem based on the results of the evaluation.
[0090] As described above, according to this embodiment, the above-mentioned ideal electron energy distribution function, evaluation function, and learning model are defined, and while measuring the electron energy distribution function of plasma during a plasma process, for example in real time, the measured electron energy distribution function of plasma can be evaluated using the evaluation function. The control device 3 can update the process parameters according to the learning model so as to approach the ideal electron energy distribution function. The control device 3 can update each coefficient in the learning model so as to improve the accuracy of the learning model according to Bayes' theorem.
[0091] Since Bayes' theorem is used, this embodiment has the advantage that the amount of calculation does not increase without bounds. In addition, this embodiment has the advantage that it is not necessary to perform a trial plasma process only for the purpose of learning, since the control law can be continuously learned through daily plasma processing.
[0092] According to this embodiment, the process parameters can be changed while measuring the electron energy distribution function of the plasma during plasma processing in real time, and the electron energy distribution function of the plasma can be controlled. Furthermore, even if the response characteristics of the plasma electron energy distribution function to changes in the process parameters are not clear, the response characteristics of the plasma electron energy distribution function to changes in the process parameters can be learned by learning based on the evaluation results of the plasma electron energy distribution function, and the control law can be optimized. Therefore, in this embodiment, it is not necessary to require high accuracy for the initial learning model that the control law setting receiving unit 204 receives from an operator or the like.
[0093] When plasma is ignited, the pressure value measured by the pressure gauge 160 may differ from the actual pressure value due to the influence of some phenomenon occurring in the processing vessel 1. Therefore, in this embodiment, when plasma is ignited, the opening degree of the pressure control valve 151 may be used instead of the pressure value measured by the pressure gauge 160 included in the process parameters. This makes it possible to obtain a more stable electron energy distribution function, and as a result, it becomes possible to realize more preferable plasma processing.
[0094] In the plasma processing of the present invention using the electron energy distribution function, ideal plasma processing can be realized in real time by replacing the ideal process result with the ideal electron energy distribution function without waiting for the measurement of the process result (measurement results such as film thickness, film quality, shape, etc. using a measurement device other than the plasma processing device) that was conventionally performed. In addition, it is an extremely useful technology because it can reduce the time required when using other measurement devices and the resources (wafers, gas, electricity, etc.) required for setting process conditions.
[0095] Although the preferred embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and various modifications and substitutions can be made to the above-described embodiments without departing from the scope of the present invention.
[0096] For example, in the present embodiment, one control device 3 corresponds to one plasma processing apparatus 100, but one control device 3 may correspond to multiple plasma processing apparatuses 100. This embodiment can also be applied to an information processing system that collects the results of learning by the control device 3 as data. In such an information processing system, a control law corresponding to, for example, the model or processing content of the plasma processing apparatus 100 can be created from the collected data, and the control law corresponding to each model or processing content can be distributed. Note that an electron energy probability function (EEPF), which corresponds to the electron energy distribution function, can also be used instead of the electron energy distribution function. Therefore, the technical content in which the electron energy distribution function is replaced with the electron energy probability function should also be included in this right.
[0097] Although the present invention has been described above based on the embodiments, the present invention is not limited to the above embodiments and various modifications are possible within the scope of the claims. This application claims priority to basic application No. 2021-152599 filed on September 17, 2021 with the Japan Patent Office, the entire contents of which are incorporated herein by reference.
[0098] For example, aspects of the present invention are as follows. <1> A plasma processing apparatus having a control device, which generates plasma from a gas supplied into a processing vessel and performs plasma processing on a workpiece, The control device includes: a measurement unit that measures an electron energy distribution function of the plasma in the processing chamber; a parameter changing unit that changes a parameter related to the plasma processing so that the electron energy distribution function measured by the measuring unit during the plasma processing approaches a target electron energy distribution function; A plasma processing apparatus comprising: <2> a target setting receiving unit that receives a setting of an ideal electron energy distribution function during the plasma processing as the target electron energy distribution function; The above further comprises <1> The plasma processing apparatus described above. <3> an evaluation function setting receiving unit that receives a setting of an evaluation function that quantifies a deviation between the electron energy distribution function measured during the plasma processing and an ideal electron energy distribution function during the plasma processing; The above further comprises <2> The plasma processing apparatus described above. <4> a control law setting receiving unit that receives a setting of a learning model that represents a correspondence between a parameter related to the plasma processing and the deviation; The above further comprises <3> The plasma processing apparatus described above. <5> The parameter changing unit changes parameters related to the plasma processing in accordance with the learning model so as to reduce the deviation. The above-mentioned <4> The plasma processing apparatus described above. <6> a control law update unit that updates the learning model based on a result of evaluation of the electron energy distribution function during the plasma processing using the changed parameters by the evaluation function; The above further comprises <4> or <5> The plasma processing apparatus described above. <7> The control law update unit updates a probability distribution of coefficient values of the learning model in accordance with Bayes' theorem based on a result of evaluation using the evaluation function. The above-mentioned <6> The plasma processing apparatus described above. <8> The parameter related to the plasma processing is at least one of an input power, a frequency of the input power, a duty ratio of an input power pulse, a gas mixture ratio, a pressure, a temperature, an accumulated number of processes, a most recent process recipe, and a most recently measured electron energy distribution function. The above-mentioned <1> ~ <7> 13. The plasma processing apparatus according to claim 12 . <9> When the plasma is ignited, the pressure included in the parameters related to the plasma processing is determined by using the opening degree of a pressure control valve instead of a pressure value measured by a pressure gauge. The above-mentioned <8> The plasma processing apparatus described above. <10> Instead of the electron energy distribution function, an electron energy probability function corresponding to the electron energy distribution function is used. The above-mentioned <1> ~ <9> 13. The plasma processing apparatus according to claim 12 . <11> A method for controlling a plasma processing apparatus having a control device, which converts a gas supplied into a processing vessel into plasma and performs plasma processing on a workpiece, comprising the steps of: The control device, measuring an electron energy distribution function of the plasma in the processing chamber; changing parameters related to the plasma processing so that the electron energy distribution function measured during the plasma processing approaches a target electron energy distribution function; The control method includes: <12> A control device controls a plasma processing apparatus that converts a gas supplied into a processing vessel into plasma and performs plasma processing on a processing target object, measuring an electron energy distribution function of the plasma in the processing vessel; changing parameters related to the plasma processing so that the electron energy distribution function measured during the plasma processing approaches a target electron energy distribution function; A program for executing the above. [Explanation of symbols]
[0099] 1 Processing vessel 2. Microwave plasma source 3. Control device 70 Plasma Probe Device 100 Plasma treatment device 200 Goal Setting Reception Department 202 Evaluation function setting reception unit 204 Control Law Setting Reception Unit 206 Plasma processing control unit 208 Electron Energy Distribution Function Measurement Section 210 Parameter change section 212 Control Law Update Section 220 Target memory section 222 Evaluation function memory section 224 Control Law Memory Unit 226 Process recipe memory unit
Claims
1. A plasma processing apparatus having a control device, which converts a gas supplied into a processing vessel into plasma and performs plasma processing on a processing object, The control device includes: a measurement unit that measures an electron energy distribution function of the plasma in the processing chamber; a parameter changing unit that changes a parameter related to the plasma processing so that the electron energy distribution function measured by the measuring unit during the plasma processing approaches a target electron energy distribution function; A plasma processing apparatus comprising:
2. a target setting receiving unit that receives a setting of an ideal electron energy distribution function during the plasma processing as the target electron energy distribution function; 2. The plasma processing apparatus according to claim 1, further comprising:
3. an evaluation function setting receiving unit that receives a setting of an evaluation function that quantifies a deviation between the electron energy distribution function measured during the plasma processing and an ideal electron energy distribution function during the plasma processing; 3. The plasma processing apparatus according to claim 2, further comprising:
4. a control law setting receiving unit that receives a setting of a learning model that represents a correspondence between a parameter related to the plasma processing and the deviation; 4. The plasma processing apparatus according to claim 3, further comprising:
5. The parameter changing unit changes parameters related to the plasma processing in accordance with the learning model so as to reduce the deviation.
5. The plasma processing apparatus according to claim 4,
6. a control law update unit that updates the learning model based on a result of evaluation of the electron energy distribution function during the plasma processing using the changed parameters by the evaluation function; 6. The plasma processing apparatus according to claim 4, further comprising:
7. The control law update unit updates a probability distribution of coefficient values of the learning model in accordance with Bayes' theorem based on a result of evaluation using the evaluation function.
7. The plasma processing apparatus according to claim 6,
8. The parameter related to the plasma processing is at least one of an input power, a frequency of the input power, a duty ratio of an input power pulse, a gas mixture ratio, a pressure, a temperature, an accumulated number of processes, a most recent process recipe, and a most recently measured electron energy distribution function.
6. The plasma processing apparatus according to claim 1, wherein:
9. When the plasma is ignited, the pressure included in the parameters related to the plasma processing is determined by using the opening degree of a pressure control valve instead of a pressure value measured by a pressure gauge.
9. The plasma processing apparatus according to claim 8,
10. Instead of the electron energy distribution function, an electron energy probability function corresponding to the electron energy distribution function is used.
6. The plasma processing apparatus according to claim 1, wherein:
11. A method for controlling a plasma processing apparatus having a control device, which generates plasma from a gas supplied into a processing vessel and performs plasma processing on a workpiece, comprising the steps of: The control device, measuring an electron energy distribution function of the plasma in the processing chamber; changing parameters related to the plasma processing so that the electron energy distribution function measured during the plasma processing approaches a target electron energy distribution function; The control method includes:
12. A control device controls a plasma processing apparatus that converts a gas supplied into a processing vessel into plasma and performs plasma processing on a processing target object, measuring an electron energy distribution function of the plasma in the processing vessel; changing parameters related to the plasma processing so that the electron energy distribution function measured during the plasma processing approaches a target electron energy distribution function; A program for executing the above.
Citation Information
Patent Citations
Calculation method of plasma corona discharge transport parameter under different humidity
CN108959706A
Simulation device and plasma discharge processing device supporting system
JP2006179407A
Film deposition system
JP2009280875A
DC / RF hybrid processing system
JP2011071522A
Simulation method, simulation program, semiconductor manufacturing apparatus
JP2013115354A
